Skip to main content

Industrialized heartbreak: how generative AI enables romance fraud at scale

Dominguez Castillo, Lorena (2026) — AI and Ethics

View Original Source (opens in a new tab)

DOI: 10.1007/s43681-026-01129-9 (opens in a new tab)

Type:   Journal Article
Country:   Global
Tags:   AI Governance, Cryptocurrency Laundering, Deepfake, Deepfakes, Fraud Detection, Generative AI, Online Safety, Platform Accountability, Romance Scams, Sextortion, Technology Facilitated Abuse
Citation tools
Download RIS Download BibTeX 


              
              

            

          

        


      

      

    

        
Transparency

        
Evidence and review status

        
This page contains AI-generated content. No human content review or subject-matter-expert review is recorded.

    


    

                    

                
Source basis

                

                    Downloaded PDF
                

            

                    

                
AI-generated page content

                

                    Yes
                

            

                    

                
Automated checks

                

                    Passed
                

            

                    

                
Administrative approval

                

                    Yes
                

            

                    

                
Human content review

                

                    Not recorded
                

            

                    

                
Subject-matter-expert review

                

                    Not recorded
                

            

            


    

        
            How this was prepared
        
        

                            

                    
Source basis

                    

                    
RSRC downloaded and privately stored a copy of the paper for internal analysis. The PDF is not offered to viewers from this page.


                                            
    
                                                            
  • 
                                    PDF added to RSRC:
                                    
                                
    • 
                                                    

                    
                                        

                

                            

                    
AI-generated page content

                    

                    
AI-generated research notes displayed on this page: Synopsis, Identified gaps, Methods, Limitations, Future work. The paper itself is not described as AI-generated.


                                            
    
                                                            
  • 
                                    Document analysis recorded:
                                    
                                
    • 
                                                            
  • 
                                    Page record updated:
                                    
                                
    • 
                                                    

                    
                                        

                

                            

                    
Automated checks

                    

                    
The current, source-bound synopsis passed the recorded versioned publication checks.


                                            
    
                                                            
  • 
                                    Checks completed:
                                    
                                
    • 
                                                    

                    
                                            

                            
                                View passed checks (3)
                            
                            
    
                                                                    
  • Length, completeness, repetition, refusal, boilerplate, and active-markup screening
    • 
                                                                    
  • Numerical claims checked against the available source text
    • 
                                                                    
  • English-source lexical grounding check
    • 
                                                            

                        

                                        

                

                            

                    
Administrative approval

                    

                    
An authenticated administrator approved the bibliographic record for public Library display. This is not a review of every research claim.


                                            
    
                                                            
  • 
                                    Approved for public display:
                                    
                                
    • 
                                                    

                    
                                        

                

                            

                    
Human content review

                    

                    
No human review is recorded for the AI-generated content displayed on this page.


                    
                                        

                

                            

                    
Subject-matter-expert review

                    

                    
RSRC has not recorded review of this content by a subject-matter or methods expert.


                    
                                        

                

                    

    


    

        Review-state definitions

        Found a possible error?
        Request a correction.
    



    



              

        
                  

            
Synopsis

            
This correspondence synthesizes empirical findings (e.g., Gressel et al.'s large-scale experiment showing 46% compliance for LLM-powered scam agents vs 18% for humans) with real-world loss data (FBI/FTC figures) and technology-trade observations on deepfakes, voice cloning, and AI-generated imagery. It identifies two converging operational models (West African Yahoo Boys and Southeast Asian compounds) and describes how LLMs preserve narrative consistency, adapt tone, and sustain dozens to hundreds of simultaneous deceptions. Fraud victims are reported to experience psychological harm comparable to violent-crime victims, including depression, anxiety, and sometimes suicidal ideation. The paper reports that romance-fraud victims are not distinguished by low intelligence or poor judgment, but may score highly on romantic idealism and addictive tendencies. The analysis relies on secondary sources and incident reports rather than primary victim data; figures reflect reported losses and may undercount true harm.

          

        
                  

            
Identified Gaps

            
The article identifies gaps in AI ethics frameworks and platform governance that neglect interpersonal AI-enabled exploitation. It notes that harms within relationships are often treated as private misfortune and not prioritized in regulation, while safety testing targets high-salience harms. It argues for a formal harm taxonomy including romance fraud and AI-facilitated abuse, enhanced fraud-training content, and robust detection of AI-generated content on platforms. Governance failures are attributed to divided responsibility among AI developers, social-media platforms, and dating services, weak accountability, under-reporting, and a lack of political constituencies to advocate for victims. An attention gap persists due to demographic targeting and stigma around reporting.

          

        
                  

            
Methods

            
This correspondence synthesizes empirical findings (e.g., Gressel et al.’s large-scale experiment showing 46% compliance for LLM-powered scam agents vs 18% for humans) with real-world loss data (FBI/FTC figures) and technology-trade observations on deepfakes, voice cloning, and AI-generated imagery. It identifies two converging operational models (West African Yahoo Boys and Southeast Asian compounds) and describes how LLMs preserve narrative consistency, adapt tone, and sustain dozens to hundreds of simultaneous deceptions. It further analyzes platform governance failures, safety-filter gaps, and the proliferation of open training materials, arguing for intervention through governance reforms, fraud-content controls, and enhanced detection across social and dating platforms.

          

        
                  

            
Limitations

            
The analysis relies on secondary sources and incident reports rather than primary victim data; figures reflect reported losses and may undercount true harm. It synthesizes studies (e.g., one major experiment on LLM-based romance baiting) that may not generalize across cultures or languages. The article acknowledges reporting biases, under-reporting by victims, and the challenge of measuring psychological harm and platform harms that are private or diffuse. It also notes safety-filter testing is incomplete, with limited visibility into open training materials and real-world abuse patterns.

          

        
                  

            
Future Work

            
Develop formal harm taxonomy for AI-enabled interpersonal abuse; study efficacy of platform-level interventions; design and evaluate fraud-detection and AI-generated-content moderation across social and dating platforms; align with EU AI Act risk-based regulation; build victim-support and reporting mechanisms; investigate cross-border enforcement and crypto-laundering links; monitor adoption of defense measures and changes in attacker tactics; foster victim-centered governance with stakeholder coalitions.

          

              

    
    

        
Related Research Insights

        

            See how this publication connects to RSRC's living evidence syntheses through current citations and research-topic mapping.